When is ‘good enough’ good enough?

The distinction every AI-age leader needs to master

👋 Hi, it's Greg and Taylor. Welcome to our newsletter on everything you wish your CEO told you about how to get ahead.

AI excels at “good enough” work.

The challenge, in the next 5 years, is going to be telling the difference between work that can stand to be “good enough” (and should be AI generated), and work that needs human intervention to be truly differentiated.

When I say “good enough,” I’m not talking about mediocre work. I’m talking about work that’s good enough to get the desired outcome (or close to it) without much or any human intervention.

If you need to write an email cadence to remind your employees to take a survey, AI can do a good enough job. Maybe it won’t get the same open rate or response rate, but it’ll get close, and it’ll save you a lot of time.

On the other hand, if you need to write a newsletter that convinces people of your subject matter expertise and drives quality lead generation, you probably need to add the human back in (although based on my LinkedIn feed these days, it seems like AI is taking over “influencer” writing).

A lot of people are going to accept “good enough” for everything they do, because it’s a lot faster and the quality looks pretty good. This will be a competitive advantage for a bit – and then a huge weakness.

Here’s how to think about “great” vs. “good enough,” and insist on great when it matters.

Greg

The framework: Great vs. good enough

Here's how I decide whether to let AI handle something or involve a human.

Does human effort create leverage on the outcome?

If improving the task by 20% won't change what happens next, let AI handle it. An email asking customers to take our feedback survey? AI is fine. Our board presentation on market expansion? That needs human thinking, because small changes in clarity could engage or disengage the board.

Is speed/volume more valuable than quality improvement?

AI's biggest advantage is speed – it can do work 1,000x faster than humans. Sometimes that speed matters more than perfect quality. At Section, we can now analyze customer conversations monthly instead of quarterly, giving us faster feedback loops even if each analysis is slightly less nuanced. The business value of 12 “good enough” analyses per year beats 4 “perfect” analyses.

What's the upside/downside of getting it wrong?

High stakes = human involvement required. Low stakes = automate it. An unengaged board costs us way more than an unoptimized expense report. A mediocre client proposal could lose us a seven-figure deal. A mediocre internal memo about PTO policy? Who cares?

Can humans actually improve the outcome by 20%?

This is the reality check. Even if quality improvement would matter to the business, can a human actually deliver it? When we tested AI vs. human instructors in our courses, completion rates only differed by 5%. The business outcome mattered, but human effort didn't move the needle enough to justify the time investment.

How much cognitive energy does the human improvement require?

This is the hidden cost. Every hour you spend perfecting a task that AI could handle at 80% quality is an hour you're not spending on work that actually differentiates you. Smart executives protect their cognitive capacity for decisions that matter.

If you need a visual reminder, this is a pretty good 4x4 to make the call.

Your new job description

The best executives in 24 months will excel at three things that weren't part of the job before:

1. Identifying where AI is good enough (and where it's not)
Not everything should be AI-generated, and not everything should be left to humans. Your job is to tell the difference.

2. Articulating the standard to your teams
You will see people on your team turning in “good enough” work (done by AI) when you need great work from them (and, conversely, people insisting on doing everything themselves because “AI can’t do what they do). As a manager, you need to articulate and enforce the difference. Say: “This project has a huge impact on our business, so we need your human brain on it.” Or alternately: “I’m fine to automate this with AI – don’t fall below this baseline, but don’t exceed it either.”

3. Managing people who do more than good enough
This is the hardest part. When AI can produce work that's genuinely competent, it becomes much harder to identify who's adding true value versus who's just refining AI output. You'll have to ask different questions: “How did you get to that conclusion? What did you consider that wasn't in the initial analysis? Where do you disagree with the AI's recommendation?”

My advice

We've always been somewhat tolerant of “good enough” work from humans. You probably have people on your team right now who consistently deliver adequate work without going above and beyond. Most of the time, we don't really notice or care.

But when someone submits AI-generated work that's merely “good enough,” our instinct is to get upset. We think they cut corners and are being lazy.

That's backwards thinking. You WANT them cutting corners with AI when “good enough” is actually good enough. You want them to automate the tasks that don't require human excellence so they can apply their cognitive energy to work that does matter.

Your job is deciding what matters – and teaching your team the same skill. I ask my executives: What's the minimum viable output, in the fastest time, for the maximum impact?

But here's the other part that matters: Every once in a while, give your team space to be craftsmen. Let them over-invest in something that probably doesn't make “productive” sense. The polish on a client deck. The attention to detail in new software. The completely rewritten blog post that's 3x better than it needed to be.

When that happens – marvel at it. Reward it somehow. In the age of AI, we need to “cut AND create.” Cut the work that can be automated, create excellence where it counts, and occasionally let people surprise you with their craft.

Have a great week,
Greg